defter*
defter / katalog / CS 559
CS 559

Deep Learning

Deep learning is fundamentally about learning hierarchical representations directly from data by training large parameterized models end-to-end with gradient descent, and this graduate course treats it as such, starting from loss surfaces and backprop, then building up to the architectures (CNNs, RNNs/attention, VAEs/GANs, deep RL) that dominate modern vision and language work. Expect a written midterm, a homework, a literature survey with a presentation, and a sizable course project where you read recent papers and implement something nontrivial in a framework like PyTorch. It assumes you are already comfortable with linear algebra, probability, and classical ML at the level of CS 464/maybe Murphy or Bishop, and it is the standard launching pad at Bilkent for thesis work in computer vision, NLP, or generative modeling.

Kredi3ECTS5FakülteFaculty of EngineeringBölümComputer EngineeringKoordinatörHamdi Dibeklioğlu

→ STARS müfredatı (resmi syllabus)

Haftalık müfredat 14 hafta

Hafta 114–20 Eyl
Machine learning, deep learning ve linear classifiers
Introduction, course structure, overview of machine learning and deep learning. Hallmarks of deep learning. Linear classifiers.
machine learningdeep learninglinear classifiers
Hafta 221–27 Eyl
Optimization: stochastic gradient descent ve back-propagation
Optimization. Stochastic gradient descent and contemporary variants, back-propagation.
stochastic gradient descentback-propagationoptimization
Hafta 328 Eyl – 4 Eki
Feedforward networks ve training teknikleri
Feedforward networks and training. Activation functions, initialization, regularization, batch normalization, model selection, ensembles.
feedforward networksactivation functionsregularizationbatch normalization
Hafta 45–11 Eki
Feedforward networks ve training teknikleri
Feedforward networks and training; Activation functions, initialization, regularization, batch normalization, model selection, ensembles
feedforward networksactivation functionsmodel selectionensembles
Hafta 512–18 Eki
CNN temelleri, pooling ve visualization
Convolutional neural networks. Fundamentals, architectures, pooling, visualization.
convolutional neural networkarchitecturepoolingvisualization
Hafta 619–25 Eki
CNN temelleri, pooling ve visualization
Convolutional neural networks. Fundamentals, architectures, pooling, visualization.
convolutional neural networkarchitecturepoolingvisualization
Hafta 726 Eki – 1 Kas
Deep learning ile spatial localization
Deep learning for spatial localization. Transposed convolution, efficient pooling, object detection, semantic segmentation.
transposed convolutionpoolingobject detectionsemantic segmentation
Hafta 82–8 Kas
RNN ve LSTM'in dil ve görü uygulamaları
Recurrent neural networks. Long-short term memory (LSTM). Language models, machine translation, image captioning, video processing, visual question answering, video processing, learning from descriptions, attention.
LSTMlanguage modelmachine translationattention
Hafta 99–15 Kas
RNN ve LSTM'in dil ve görü uygulamaları
Recurrent neural networks. Long-short term memory (LSTM). Language models, machine translation, image captioning, video processing, visual question answering, video processing, learning from descriptions, attention.
LSTMlanguage modelmachine translationattention
Hafta 1016–22 Kas
RNN ve LSTM'in dil ve görü uygulamaları
Recurrent neural networks. Long-short term memory (LSTM). Language models, machine translation, image captioning, video processing, visual question answering, video processing, learning from descriptions, attention.
LSTMlanguage modelmachine translationattention
Hafta 1123–29 Kas
Deep generative models ve representation learning
Deep generative models. Auto-encoders, variational auto-encoders, generative adversarial networks, auto-regressive models, generative image models, unsupervised and self-supervised representation learning.
variational auto-encodersgenerative adversarial networksauto-regressive modelsself-supervised representation learning
Hafta 1230 Kas – 6 Ara
Deep generative models ve representation learning
Deep generative models. Auto-encoders, variational auto-encoders, generative adversarial networks, auto-regressive models, generative image models, unsupervised and self-supervised representation learning.
variational auto-encodersgenerative adversarial networksauto-regressive modelsself-supervised representation learning
Hafta 137–13 Ara
Deep reinforcement learning yöntemleri
Deep reinforcement learning. Policy gradient methods, Q-Learning.
deep reinforcement learningpolicy gradientQ-Learning
Hafta 1414–20 Ara
Deep reinforcement learning ve proje sunumları
Deep reinforcement learning. Project presentations.
deep reinforcement learningproje sunumu

Değerlendirme 100% · 4 adım

30%
20%
15%
35%
Midterm:Essay/written Midterm 30%
Homework Homework 20%
Literature Survey & Presentation Literature Survey & Presentation 15%
Project Project 35%
en büyük tek kalem %35 · sınav ağırlığı %30 · 9 dönem ortalaması 3.49 (305 öğrenci) nasıl hesaplanıyor

Önerilen kaynaklar 3 kitap

📖
Önerilen
Deep Learning
I. Goodfellow, Y. Bengio
A. Courville · 2016
📖
Önerilen
Machine Learning: A Probabilistic Perspective
K. P. Murphy
2012 · MIT Press
📖
Önerilen
Pattern Recognition and Machine Learning
C. M. Bishop
2006 · Springer

Bu dersi alınca · 8 öğrenme çıktısı

Bilkent'in resmî syllabus'ünden. Sağdaki etiket o çıktının hangi değerlendirmeyle ölçüldüğünü söylüyor.

Ders notları · henüz yok

CS 559 için defter ekibi henüz not yazmadı.

İlk dosyayı sen atarsan: not, slayt, geçmiş sınav, çözüm, cheat-sheet, ne varsa. defter ekibi öğrenci paylaşımlarından bu dersin notlarını yazar. Drive linki / PDF / ZIP, hepsi olur.

← katalog

Geçmiş GPA dağılımı 9 dönem · ort. 3.49

DönemCourse CPA
2025-2026 Fall 3.59 1 sec · 40 öğr
2024-2025 Fall 3.42 1 sec · 40 öğr
2023-2024 Fall 3.59 1 sec · 41 öğr
2022-2023 Fall 3.16 1 sec · 27 öğr
2021-2022 Fall 3.48 1 sec · 17 öğr
2020-2021 Spring 3.40 1 sec · 39 öğr
2019-2020 Spring 3.55 1 sec · 39 öğr
2018-2019 Spring 3.56 1 sec · 35 öğr
2016-2017 Spring 3.62 1 sec · 27 öğr

Aggregate course GPA · Bilkent STARS'tan public data. Hoca-bazlı per-section detayı için STARS evaluation report →. Öğrenci anket cevapları KVKK kapsamında defter'de tutulmaz. Tüm derslerin ortalamaları →

2026-2027 Güz için şubesi henüz görünmüyor. Kayıt sistemi bu dersi bu dönem listelemiyor, ama şube girişi sürüyor: ders kaydı 15 Eylül, bölümler o güne kadar şube ekleyebiliyor. Son 4 güz döneminin hepsinde açılmış, yani bu dönem de açılması beklenir. Son açıldığı dönemde (2025-2026 güz) 1 şube vardı. Ön kayıt müfredat üzerinden yapılıyor, açılan şube listesi üzerinden değil; o yüzden ön kayıtta seçebildiğin bir dersin şubesi burada henüz görünmeyebilir. Kesin sonuç ders kaydında belli oluyor. kayıt tarihleri → · açık dersler

⚠️ FZ engelleyen şartlar

There is no final exam for this course, however, any one of the following will directly result in an F grade: (1) not submitting a project or homework (including report), (2) not preparing/presenting a survey on the pre-scheduled date, (3) being absent in the midterm, (4) being absent in a project presentation.

Hocalar 0 bu dönem · 2 geçmiş

Geçmişte ders veren (2 kişi)
Hamdi Dibeklioğlu, Ramazan Gökberk Cinbiş